基于边缘计算和改进YOLOv5s算法的输电线路故障实时检测方法

doi:10.12204/j.issn.1000-7229.2023.01.011

电力建设 ›› 2023, Vol. 44 ›› Issue (1): 91-99.doi: 10.12204/j.issn.1000-7229.2023.01.011

基于边缘计算和改进YOLOv5s算法的输电线路故障实时检测方法

黄悦华(), 陈照源(), 陈庆(), 张磊(), 刘恒冲(), 张家瑞()

三峡大学电气与新能源学院,湖北省宜昌市 443002

收稿日期:2022-04-08 出版日期:2023-01-01 发布日期:2022-12-26
通讯作者: 陈庆 E-mail:hyh@ctgu.edu.cn;897201539@qq.com;chenqing20190808@163.com;leizhang3188@163.com;2162015559@qq.com;892676757@qq.com
作者简介:黄悦华(1972),男,博士,教授,主要研究方向为智能电网、新能源微电网、综合能源系统,E-mail:hyh@ctgu.edu.cn。
陈照源(1995),男,硕士研究生,主要研究方向为电力边缘智能和电力深度视觉,E-mail:897201539@qq.com。
张磊(1986),男,博士,副教授,主要研究方向为人工智能技术在电力系统的应用,E-mail:leizhang3188@163.com。
刘恒冲(1998),男,硕士研究生,主要研究方向为输变电设备故障诊断和图像处理,E-mail:2162015559@qq.com。
张家瑞(1994),男,硕士研究生,主要研究方向为人工智能技术在电力系统的应用,E-mail:892676757@qq.com。
基金资助:
国家自然科学基金项目(52007103);湖北省科技重大专项(2020AEA012)

Real-Time Detection Method for Transmission Line Faults Applying Edge Computing and Improved YOLOv5s Algorithm

HUANG Yuehua(), CHEN Zhaoyuan(), CHEN Qing(), ZHANG Lei(), LIU Hengchong(), ZHANG Jiarui()

College of Electrical Engineering and New Energy,China Three Gorges University,Yichang 443002,Hubei Province,China

Received:2022-04-08 Online:2023-01-01 Published:2022-12-26
Contact: CHEN Qing E-mail:hyh@ctgu.edu.cn;897201539@qq.com;chenqing20190808@163.com;leizhang3188@163.com;2162015559@qq.com;892676757@qq.com

摘要/Abstract

摘要：

随着输电线路无人机巡检工作的常态化,暴露出故障图像检测实时性、模糊目标检测精准性难以满足实际工作需求的问题。文章提出一种基于边缘计算和改进YOLOv5s算法的输电线路故障实时检测方法。以YOLOv5s为基础检测模型,基于Ghost轻量化模块重构模型获取数据特征的卷积操作过程,提高了模型的检测速度;采用基于KL散度分布的损失函数作为目标框定位损失函数,提升了模型对模糊图像检测的精度。将改进的YOLOv5s算法部署于华为Atlas 200 DK边缘模块中,对绝缘子自爆、防震锤脱落、鸟巢3类故障进行检测,其平均精度均值可达84.75%,检测速度为34 frame/s。结果表明,改进的算法在保证检测实时性的同时,能够提升对模糊故障目标图像的检测精度,满足无人机搭载边缘设备的输电线路巡检需求。

关键词: 边缘计算, 输电线路故障, YOLOv5s, 实时检测, Ghost轻量化模块, KL散度

Abstract:

With the normalization of unmanned aerial vehicle (UAV) inspection of transmission lines, the problems of real-time detection of fault images and accuracy of blurred target detection are difficult to meet the actual work requirements. This paper proposes a real-time detection method for transmission line faults, which is based on edge computing and improved YOLOv5s algorithm. This method is based on YOLOv5s model, and the model is reconstructed on the basis of Ghost lightweight module to realize the convolution operation process of obtaining data features, which improves the detection speed of the model. The loss function based on KL (Kullback-Leibler) divergence distribution is used as the target box localization loss function to improve the accuracy of blurred image detection. The improved YOLOv5s algorithm is deployed in the Huawei Atlas 200 DK edge module to detect three types of faults: insulator self-explosion, shock hammer falling-off, and bird’s nest. The mAP can reach 84.75%, and the detection speed is 34 frame/s. The results show that the improved algorithm can improve the detection accuracy of blurred fault target images while ensuring the real-time detection, and meet the inspection requirements of transmission lines equipped with edge devices by UAV.

This work is supported by National Natural Science Foundation of China (No. 52007103) and Major Science and Technology Projects in Hubei Province of China (No. 2020AEA012).

Key words: edge computing, transmission line fault, YOLOv5s, real-time detection, Ghsot lightweight module, KL divergence

中图分类号:

TM75

黄悦华, 陈照源, 陈庆, 张磊, 刘恒冲, 张家瑞. 基于边缘计算和改进YOLOv5s算法的输电线路故障实时检测方法[J]. 电力建设, 2023, 44(1): 91-99.

HUANG Yuehua, CHEN Zhaoyuan, CHEN Qing, ZHANG Lei, LIU Hengchong, ZHANG Jiarui. Real-Time Detection Method for Transmission Line Faults Applying Edge Computing and Improved YOLOv5s Algorithm[J]. ELECTRIC POWER CONSTRUCTION, 2023, 44(1): 91-99.

图/表 14

图1 基于边缘计算的输电线路故障实时检测框架

Fig.1 Real-time detection framework for transmission line faults applying edge computing

图2 YOLOv5s模型结构

Fig.2 YOLOv5s model structure

图3 特征图可视化

Fig.3 Visualization of feature map

图4 Conv卷积生成特征过程

Fig.4 Conv to generate feature process

图5 Ghost原理示意图

Fig.5 Principle diagram of Ghost

图6 数据集图示

Fig.6 Data set diagram

表1 数据集分布

Table 1 Distribution of the data set

图7 边缘计算设备

Fig.7 Edge computing device

图8 Loss曲线

Fig.8 Loss curve

图9 mAP曲线

Fig.9 mAP curve

图10 F1曲线

Fig.10 F1 curve

表2 不同改进策略的对比实验

Table 2 Comparative experiment of different improvement strategies

图11 模糊故障图像检测结果

Fig.11 Detection results of blurred fault image

表3 主流模型性能对比

Table 3 Performance comparison of mainstream models

参考文献 20

[1]	戴彦, 王刘旺, 李媛, 等. 新一代人工智能在智能电网中的应用研究综述[J]. 电力建设, 2018, 39(10): 1-11. doi: 10.3969/j.issn.1000-7229.2018.10.001
	DAI Yan, WANG Liuwang, LI Yuan, et al. A brief survey on applications of new generation artificial intelligence in smart grids[J]. Electric Power Construction, 2018, 39(10): 1-11. doi: 10.3969/j.issn.1000-7229.2018.10.001
[2]	DENG C, WANG S W, HUANG Z, et al. Unmanned aerial vehicles for power line inspection: a cooperative way in platforms and communications[J]. Journal of Communications, 2014, 9(9): 687-692. doi: 10.12720/jcm.9.9.687-692 URL
[3]	马富齐, 王波, 董旭柱, 等. 电力视觉边缘智能:边缘计算驱动下的电力深度视觉加速技术[J]. 电网技术, 2020, 44(6): 2020-2029.
	MA Fuqi, WANG Bo, DONG Xuzhu, et al. Power vision edge intelligence: power depth vision acceleration technology driven by edge computing[J]. Power System Technology, 2020, 44(6): 2020-2029.
[4]	许德刚, 王露, 李凡. 深度学习的典型目标检测算法研究综述[J]. 计算机工程与应用, 2021, 57(8): 10-25. doi: 10.3778/j.issn.1002-8331.2012-0449
	XU Degang, WANG Lu, LI Fan. Review of typical object detection algorithms for deep learning[J]. Computer Engineering and Applications, 2021, 57(8): 10-25. doi: 10.3778/j.issn.1002-8331.2012-0449
[5]	REDMON J, FARHADI A. YOLOv3: an incremental improvement[EB/OL]. [2021-10-18]. https://arxiv.org/abs/1804.02767.
[6]	BOCHKOVSKIY A, WANG C Y, LIAO H Y M. YOLOv4: optimal speed and accuracy of object detection[EB/OL]. [2021-10-18].https://doi.org/10.48550/arXiv.2004.10934.
[7]	Ultralytics. YOLOv5[EB/OL].(2020-06-09)[2021-10-18]. ttps://github.com/ultralytics/yolov5.
[8]	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[C]// Computer Vision - ECCV 2016. Cham: Springer International Publishing, 2016: 21-37.
[9]	陈文贺, 李彩林, 袁斌, 等. 有效的绝缘子自爆缺陷定位方法[J]. 计算机工程与设计, 2019, 40(8): 2346-2352.
	CHEN Wenhe, LI Cailin, YUAN Bin, et al. Effective method to locate self-explosion defects of insulators[J]. Computer Engineering and Design, 2019, 40(8): 2346-2352.
[10]	祁婕, 焦良葆. 改进SSD的输电铁塔鸟窝检测[J]. 计算机系统应用, 2020, 29(5): 202-208.
	QI Jie, JIAO Liangbao. Bird nest detection on transmission tower based on improved SSD algorithm[J]. Computer Systems & Applications, 2020, 29(5): 202-208.
[11]	刘悦, 黄新波. 基于YOLOv4和改进分水岭算法的绝缘子爆裂检测定位研究[J]. 电网与清洁能源, 2021, 37(7): 51-57.
	LIU Yue, HUANG Xinbo. Research on insulator burst fault identification based on YOLOv4 and improved watershed algorithm[J]. Power System and Clean Energy, 2021, 37(7): 51-57.
[12]	YANG Q F, ZHANG Z Q, YAN L, et al. Lightweight bird's nest location recognition method based on YOLOv4-tiny[C]//2021 IEEE International Conference on Electrical Engineering and Mechatronics Technology. Qingdao, China. IEEE, 2021 : 402-405.
[13]	翁智, 程曦, 郑志强. 基于改进YOLOv3的高压输电线路关键部件检测方法[J]. 计算机应用, 2020, 40(S2): 183-187.
	WENG Zhi, CHENG Xi, ZHENG Zhiqiang. Detection method of key components of high-voltage transmission lines based on improved YOLOv3[J]. Journal of Computer Applications, 2020, 40(S2): 183-187.
[14]	ZHANG S, CHEN B, WANG R S, et al. Unmanned aerial vehicle (UAV) vision-based detection of power line poles by CPU-based deep learning method[C]//2019 IEEE 9th Annual International Conference on CYBER Technology in Automation, Control, and Intelligent Systems. Suzhou, China. IEEE, 2019: 1630-1634.
[15]	李一鸣, 王潇. 基于YOLOv5s模型的轧钢表面缺陷检测[J]. 制造业自动化, 2021, 43(11): 117-119.
	LI Yiming, WANG Xiao. Surface defect detection of rolled steel based on YOLOv5s model[J]. Manufacturing Automation, 2021, 43(11): 117-119.
[16]	蒋润熙, 阿里甫·库尔班, 耿丽婷. 面向轻量化网络的安全帽检测算法[J]. 计算机工程与应用, 2021, 57(20): 263-270. doi: 10.3778/j.issn.1002-8331.2105-0347
	JIANG Runxi, ALIFU Kuerban, GENG Liting. Safety helmet detection algorithm for lightweight network[J]. Computer Engineering and Applications, 2021, 57(20): 263-270. doi: 10.3778/j.issn.1002-8331.2105-0347
[17]	TRINH H C, LE D H, KWON Y K. PANET: a GPU-based tool for fast parallel analysis of robustness dynamics and feed-forward/feedback loop structures in large-scale biological networks[J]. PLoS One, 2014, 9(7): e103010. doi: 10.1371/journal.pone.0103010 URL
[18]	HAN K, WANG Y H, TIAN Q, et al. GhostNet: more features from cheap operations[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle, WA, USA. IEEE, 2020 : 1577-1586.
[19]	尹丽华, 李范鸣, 刘士建. 基于自适应补偿的电子稳像方法[J]. 激光与红外, 2017, 47(11): 1438-1445.
	YIN Lihua, LI Fanming, LIU Shijian. Electronic image stabilization method based on adaptive compensation[J]. Laser & Infrared, 2017, 47(11): 1438-1445.
[20]	HE Y H, ZHU C C, WANG J R, et al. Bounding box regression with uncertainty for accurate object detection[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach, CA, USA. IEEE, 2019: 2883-2892.

基于边缘计算和改进YOLOv5s算法的输电线路故障实时检测方法

Real-Time Detection Method for Transmission Line Faults Applying Edge Computing and Improved YOLOv5s Algorithm

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 14

参考文献 20

相关文章 11

编辑推荐

Metrics

本文评价

[1]	秦强, 刘文泽, 谭炜豪, 蔡泽祥, 岑伯维, 匡佩. 面向软件定义网络的配电边缘计算终端优化部署方法[J]. 电力建设, 2023, 44(1): 82-90.
[2]	杨立滨, 张磊, 刘艳章, 李正曦, 宗鸣. 基于分布式框架的新能源场站并网性能评估[J]. 电力建设, 2022, 43(5): 137-144.
[3]	潘思宇, 刘宝柱. 面向配电网故障快速处理的边缘计算单元优化配置方法[J]. 电力建设, 2022, 43(3): 31-41.
[4]	王同贺, 华昊辰, 曹军威. 共识边缘计算及其在能源互联网中的应用[J]. 电力建设, 2021, 42(2): 116-125.
[5]	原吕泽芮，顾洁，金之俭. 基于云-边-端协同的电力物联网用户侧数据应用框架[J]. 电力建设, 2020, 41(7): 1-8.
[6]	徐晓，陈中，丁宏恩，方国权，陈妍希，杜璞良. 面向区域售电公司的边缘计算架构设计探讨[J]. 电力建设, 2019, 40(7): 41-47.
[7]	杨军，林洋佳，陈杰军，李逸欣，詹祥澎. 未来城市共享电动汽车发展模式[J]. 电力建设, 2019, 40(4): 49-59.
[8]	祁兵1,2, 夏琰1, 李彬1, 石坤2, 薛溟枫3. 基于边缘计算的家庭能源管理系统：架构、关键技术及实现方式[J]. 电力建设, 2018, 39(3): 33-.
[9]	刘思放，邓春宇，张国宾，祁兵，李彬，李德智，石坤，杨斌，奚培锋. 面向居民智能用电的边缘计算协同架构研究[J]. 电力建设, 2018, 39(11): 60-68.
[10]	罗杰，蒋铁铮，梁勇超，马瑞. 500 kV输电线路故障时空特性分析[J]. 电力建设, 2015, 36(6): 46-52.
[11]	戴文进,郭世才 . 高压输电线路故障的红外诊断技术[J]. 电力建设, 2005, 26(4): 46-0.